glmtools

GOES Geostationary Lightning Mapper Tools

Installation

glmtools requires Python 3.5+ and provides a conda environment.yml for the key dependencies.

See the documentation in docs/index.rst for complete installation instructions.

Description

Compatible data:

NetCDF format Level 2 data, as described in the Product Definition and Users Guide

glmtools automatically reconstitutes the parent-child relationships implicit in the L2 GLM data and adds traversal information to the dataset:

calculating the parent flash id for each event
calculating the number of groups and events in each flash
calculating the number of events in each group

xarray's dimension-aware indexing lets you quickly reduce the dataset to flashes of interest, as described below.

glmtools can restore the GLM event geometry using a built-in corner-point lookup table, which allows for gridding of the imagery at finer resolutions that accurately represent the full footprint of each event, group, and flash.

Some common tasks

Create gridded NetCDF imagery

Use the script in examples/grid/make_GLM_grids.py. For instance, the following command will grid one minute of data (3 GLM files) on the ABI fixed grid in the CONUS sector at 2 km resolution. These images will overlay precisely on the ABI cloud tops, and will have parallax with respect to ground for all the same reasons ABI does.

python make_GLM_grids.py -o /path/to/output/ --fixed_grid --split_events \
--goes_position east --goes_sector conus --dx=2.0 --dy=2.0 --ctr_lon 0.0 --ctr_lat 0.0 \
--start=2018-01-04T05:37:00 --end=2018-01-04T05:38:00 \
OR_GLM-L2-LCFA_G16_s20180040537000_e20180040537200_c20180040537226.nc \
OR_GLM-L2-LCFA_G16_s20180040537200_e20180040537400_c20180040537419.nc \
OR_GLM-L2-LCFA_G16_s20180040537400_e20180040538000_c20180040538022.nc \

To start with, look at the flash extent density and total energy grids.

ctr_lon and ctr_lat aren't used, but are required anyway. Fixing this would make a nice first contribution!

Removing the --split_events flag and setting the grid to 10 km allows for gridding of the raw point data, and will run much faster. Finer resolutions will cause gaps in flash extent density because the point data are spaced about 8-12 km apart. Note that these grids will either have gaps or will double-count events along GLM pixel borders, because there is no one grid resolution which exactly matches the GLM pixel size as it varies with earth distortion over the field of view.

Interactively plot raw flash data

See the examples folder. plot_glm_test_data.ipynb is a good place to start.

Reduce the dataset to a few flashes

from glmtools.io.glm import GLMDataset
filename = 'OR_GLM-L2-LCFA_G16_s20180040537000_e20180040537200_c20180040537226.nc'
glm =  GLMDataset(filename)
flash_id_list = glm.dataset.flash_id[20:30]
smaller_dataset = glm.get_flashes(flash_id_list)

Filter out flashes geographically or by events/groups per flash

See glmtools.io.glm.GLMDataset.subset_flashes.

The logic implemented above is pretty simple, and below shows how to adapt it to find large flashes.

from glmtools.io.glm import GLMDataset
filename = 'OR_GLM-L2-LCFA_G16_s20180040537000_e20180040537200_c20180040537226.nc'
glm =  GLMDataset(filename)
fl_idx = glm.dataset['flash_area'] > 2000
flash_ids = glm.dataset[{glm.fl_dim: fl_idx}].flash_id.data
smaller_dataset = glm.get_flashes(flash_ids)
print(smaller_dataset)

bmatilla/glmtools